Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
This paper reviews recent progress on understanding the effects of compliant walls on boundary-layer instabilities and laminar-turbulent transition. Aspects of the structure and properties of dolphin skin are also considered insofar as they have influenced the work on the hydrodynamics of compliant walls. The question of whether the work on this subject can shed any light on the supposed laminar-flow capability of dolphin skin is also examined. DO some dolphin species (e.g. the bottle-nosed dolphin Tursiops truncatus) possess an extraordinary laminarflow capability? Certainly mankind has long admired the swimming skills of this fleet creature. Scientific interest in dolphin hydrodynamics dates back at least as far as 1936 when Gray1 published his analysis of dolphin energetics. It is widely accepted (although there is expert dissenting opinion2,3) that Tursiops truncatus can maintain a sustained swimming speed of 9 m/s. Gray followed the usual practice of marine engineers and modelled the dolphins body as a one-sided flat plate of length 2 m. The corresponding value of Reynolds number, ReL, based on the body length was about 20 10 6. Even in a very low-noise flow environment the Reynolds number, Rext, for transition from laminar to turbulent flow does not exceed 2 to 3 10 6 for flow over a flat plate. Accordingly, Gray assumed that if conventional hydrodynamics were involved, the flow would mostly be turbulent and the dolphin body would experience a large drag force. So large, in fact, that at 9 m/s its muscles would have to deliver about seven times more power per unit mass than any other mammalian muscle. This led him and others to argue that the dolphin must be capable of maintaining laminar flow by some extraordinary means. This hypothesis has come to be known as Grays Paradox. Little in detail was known about laminar-turbulent transition in 1936 and Gray would have been unaware of the effect of the streamwise pressure gradient exter For correspondence. (e-mail: pwc@eng.warwick.ac.uk) The first author does not know Prof. Dhawan well, but has had the pleasure of hearing him lecture on two occasions. We hope that our paper which combines stability and transition with dolphin hydrodynamics will serve as a modest tribute to his work.
nal to the boundary layer. We now know that transition is delayed in favourable pressure gradients (accelerating flow) and promoted by adverse ones (decelerating flow)4. Thus, for the dolphin, the transition point would be expected to occur near the point of minimum pressure. For Tursiops truncatus this occurs about half way along the body5 corresponding to Rext = 10 10 6. When this is taken into account the drag is very much less and the required power output from the muscles only exceeds the mammalian norm by no more than a factor of two. There is also more recent evidence that dolphin muscle is capable of higher output2,3. So after reexamination of Grays paradox there is now much less to explain. Nevertheless, the dolphin may still find advantage in a laminar-flow capability. Moreover, there is ample evidence, which we will review in the present paper, that the use of passive artificial dolphin skins compliant walls can maintain laminar flow. We will also review the information on those features of the dolphin epidermis which may be of hydrodynamics significance. In particular, we will consider whether the study of compliant walls can offer any evidence that dolphin skin has laminar-flow properties.
758
Figure 1. Kramers compliant coating and model. All dimensions in mm. (Drawings based on those given in ref. 7.) a, Cross-section; b, Cut through stubs; c, Model: shaded regions were coated. (Based on figure 1 of ref. 17.)
(c)
b
Figure 2. Structure of Dolphin skin. a, Cross-section; b, Cut through dermal papillae at AA'; c, Front view. Key: a, cutaneous ridges or microscales; b, dermal papillae; c, dermal ridge; d, upper epidermal layer; e, fatty tissue.
running approximately normal to the flow direction. The effect of the cutaneous ridges on the hydrodynamics is unknown. The upper epidermal layer forms a comparatively dense elastic membrane and is considCURRENT SCIENCE, VOL. 79, NO. 6, 25 SEPTEMBER 2000
SPECIAL SECTION:
coatings acted as damping to suppress the growth of the waves. This must have seemed eminently reasonable at the time. Surprisingly, however, the early theoretical work by Benjamin14,15 and Landahl16, while showing that wall compliance can indeed suppress the growth of T/S waves, also showed that wall damping itself promoted wave growth (i.e. the waves grew faster for a high level of damping than for a low level). Because the theory contradicted his concept of the role of damping and also owing to the failure of other investigators to confirm his results, few believed that Kramers coating had a laminar-flow capability. But Benjamin and Landahls theory was rather general and they had made no attempt to model Kramers coatings theoretically. A theoretical assessment of his coatings was carried out much later by Carpenter and Garrad17,18 who modelled the coatings as plates supported on a spring foundation with the effects of visco-elastic damping and the viscous damping fluid included. Their results broadly confirmed that the Kramer coatings were capable of substantially reducing the growth of T/S waves. Experimental confirmation was still lacking for the stabilizing effects of wall compliance on T/S waves. This was provided by Gaster19. A schematic diagram of his compliant panel and experimental set-up is shown in Figure 3. The compliant wall was simpler than Kramers and less obviously like dolphin skin. It consisted of two-layers: a thin, outer, plate-like covering surmounting a much softer and thicker layer. The test model was a flat plate with a compliant-panel insert. The T/S waves were created by a driver located ahead of the compliant panels leading edge and they were measured at its trailing edge by a surface hot-film foil gauge. Close agreement was found between the measured growth and that predicted by suitably modified linear stability theory. seminal papers of Benjamin14,15 and Landahl16. Benjamins discovery of the unexpected effect of wall damping on the T/S waves, led him and Landahl to introduce a new concept to science. They classified waves according to how they responded to irreversible energy transfer. Class A waves (now called Negative Energy Waves NEW after terminology introduced later in plasma dynamics22) are destabilized (stabilized) by irreversible energy transfer out of (into) the system due, for example, to wall damping. T/S waves belong to this class. (One needs to be a little cautious in this regard as strictly the concept can only be applied to conservative base states.) Class B waves (Positive Energy Waves PEW) are more conventional in that they are stabilized (destabilized) by energy transfer processes having the opposite effect on the NEW. In their pure forms the NEW and PEW can only become convective instabilities22, i.e. they grow exponentially (until nonlinear effects intervene) as they propagate downstream, but do not grow with time at a fixed location. Owing to their opposite energy requirements, however, NEW and PEW can combine to form a truly self-sustained, temporallygrowing instability, known as an absolute instability22. Once they form, these instabilities are indifferent to irreversible energy transfer. Examples of such instabilities found in flow over compliant walls are discussed below. An interesting, and highly significant, observation made in Gasters experimental study was that for the two most compliant of his three panels, the route to transition was not amplification of T/S waves. Unlike the relatively gradual process found for the rigid control, transition occurred suddenly when a critical speed was reached. Moreover, when this happened the signal from the hot-film foil gauge located at the panels trailing edge oscillated at a much higher frequency than the driver. It was later shown by Lucey and Carpenter23 that a flow-induced surface instability traveling-wave flutter set in at the observed transition speed. (This instability is included schematically in Figure 3). Traveling-wave flutter is a PEW and it is destabilized by irreversible energy transfer to the wall due to the work done on it by the fluctuating pressure. If the boundary layer were absent it would be found that the fluctuating pressure generated in an unsteady potential flow when a surface wave propagates along a compliant surface is exactly 90 degrees out of phase with the vertical velocity of the wall. Thus no network would be transferred over the wave period. If the boundary layer can shift this phase difference away from 90 degrees then there would be a possibility of irreversible energy transfer to the wall. Benjamin15 showed that a mechanism originally identified by Miles in connection with water waves could also apply to waves on compliant walls. He showed that the required phase change in pressure occurred at the critical point where the local
CURRENT SCIENCE, VOL. 79, NO. 6, 25 SEPTEMBER 2000
velocity in the boundary layer equals the wave speed. Thus, although viscous effects were required to create the boundary layer, the mechanism is otherwise inviscid. A knowledge of this mechanism allowed simple and accurate estimates to be made for the critical wavenumber and flow speed of traveling-wave flutter in the case of the plate-spring model18. A more rigorous analysis based on asymptotic techniques and including other irreversible mechanisms was developed by Carpenter and Gajjar24. This was later extended to the more complex case of Gaster-type two-layer walls21,23. To summarize the attributes of traveling-wave flutter: it is a convective instability that travels at speeds of the order of 0.7 U ; it is stabilized by irreversible energy transfer to the wall, whereas energy transfer out of the wall, such as damping, has the opposite effect and can be used to control it. It is much more sensitive to wall damping than T/S waves and it appears that the true role of the damping fluid in Kramers coatings was to control traveling-wave flutter17. Figure 4 presents the results of a numerical simulation of traveling-wave flutter propagating in the boundary layer over a finite compliant panel of plate-spring type. The figure shows ray paths traced out by a propagating wave-packet initiated by creating a small bump on the left of the domain which is then allowed to relax. It can be seen that, in fact, two separate wave-packets are present. Despite appearances, the right-hand one becomes dominant at later times. The computations were carried out using our novel discrete-vortex method25,26. Traveling wave flutter is now well understood and can be confidently predicted with existing theory. The other commonly observed flow-induced surface instability is divergence. This is a simple instability to understand. Imagine a small disturbance in the form of a bump is somehow created on a compliant surface. There will be a pressure drop as the flow passes over the bump, thereby creating a suction force. If the flow speed is steadily increased this suction force will rise and at a sufficiently high flow speed it will outweigh the restorative structural force in the wall causing the
CURRENT SCIENCE, VOL. 79, NO. 6, 25 SEPTEMBER 2000
bump to grow until checked by the rise in the structural force due to nonlinear effects. Thus the physical mechanism is conservative and does not require any viscous effects at all; it can occur even in a potential flow27. In fact, simple estimates for the critical wave-number and flow speed were derived for potential flow over the plate-spring compliant-wall model28. These appeared to give reasonable agreement with the experimental data then available. According to this model, divergence is a static wave at the point of instability and slowly travels downstream at supercritical flow speeds27,28. It has been observed on the underside and rear of dolphins when they swim at high speed for short durations5,29. It has also been seen on the original Kramer compliant coatings30. The phenomenon is not quite so straightforward for the Gaster-type, two-layer walls and for the simpler one-layer walls (Gaster walls with the top plate-like layer removed). For example, Duncan et al.31 found that the critical wavelength was infinitesimally small for the single-layer walls. Divergence on such walls was investigated in detail in a seminal experimental study by Gad-el-Hak et al.32. They found that the divergence waves traveled slowly downstream at speeds between 0.02 U and 0.05 U . The waves exhibited sharp crests with broad valleys between each wave. Perhaps the most significant finding was that the divergence waves only occurred when the flow was turbulent. This is most dramatically illustrated in their figure 10 which showed a wedge of turbulence created by a local roughness element surrounded by laminar flow. The divergence waves were only found within the turbulent wedge. Duncan et al. explained that the presence of the shear flow in the boundary layer reduced the magnitude of the pressure fluctuations. This shear-sheltering effect would have to be much greater for laminar boundary layers compared with turbulent ones in order to explain the observations of Gad-el-Hak et al. In a subsequent experimental study using an ultra-low-damping material, Gad-el-Hak33 was also able to produce traveling wave flutter on a single-layer wall, but not divergence in this
761
SPECIAL SECTION:
technique was used and, as before, the disturbance is created by a relaxing bump, this time located in the centre of the domain. The main difference between the two figures is that the flow conditions are now such that the flow/wall system is unstable with respect to divergence. It can be seen that the ray patterns and displacements are quite different from Figure 4. Now the wave-packet does not propagate downstream but spreads in both directions from the point of initiation. This is characteristic of an absolute instability. In fact, Yeo et al.34 have shown rigorously that divergence is an absolute instability. By means of our numerical simulations we have been able to confirm that the pressure fluctuations are reduced to a much greater extent in a laminar boundary layer than for a comparable turbulent one26 owing to the smaller shear thickness of the latter. The peculiar shape of the divergence waves revealed in the experiments of Gad-el-Hak et al.32 has recently been explained by Lucey et al.35. They showed that it is necessary to include nonlinear effects in both the fluid and wall dynamics. When this was done the hydrodynamic stiffness became increasingly peaky, similar to the wave-forms observed in the experiments, as the wave amplitude increased. Under certain circumstances the NEW T/S waves can coalesce with the PEW traveling-wave flutter instability to form a much more powerful instability36 which has been termed a transitional mode by Sen and Arora37. It appears that excessive use of wall damping to control traveling-wave flutter can give rise to this instability17. Thus it sets an upper limit on the level of damping that can be used to control traveling-wave flutter. But wall damping is not essential for its existence. The group velocity is zero and it appears to have the attributes of an absolute instability. Davies and Carpenter38 have shown that this transitional mode replaces divergence for laminar plane channel flow. It is now known that this also happens for the flat-plate boundary layer39. In practice, an experimentalist would find it difficult to distinguish between divergence and the transitional mode as they are both absolute instabilities. Compliant walls can also support evanescent modes20,22,40. These are spatially-developing, attenuating waves which usually have the phase and group velocities in opposite directions. Figure 6 displays a numerical simulation of a T/S wave which is incident on a finite compliant panel. The spectrum in Figure 6 a shows that two other modes are present over the compliant panel as well as the T/S wave. These are both convective eigenmodes of the coupled boundary-layer/ compliant-wall system. One mode has a similar wavelength to the T/S wave and propagates upstream from the panels trailing edge. The other is a lightly damped, near-neutral, much longer, wave which propagates downstream from the panels leading edge. Some evanescent modes may also coalesce with other waves to form powerful absolute instabilities27,34. A good examCURRENT SCIENCE, VOL. 79, NO. 6, 25 SEPTEMBER 2000
Figure 4. Wall displacement profiles obtained by numerical simulation of traveling-wave flutter produced by a relaxing bump located near the leading edge of the compliant panel. The flow is from left to right. (Based on figure 2 b of ref. 26.)
Figure 5. Numerical simulation of divergence: a, space-time form; b, wall displacement profiles at successive times. (Based on figure 4 of ref. 26.)
case. In contrast to the previous case, the wave-forms were much closer to the sinusoid form and traveled at around 0.5 U . Figure 5 presents a numerical simulation of divergence which can be compared with Figure 4. The same
762
SPECIAL SECTION:
tailored to the local flow environment. (Short compliant panels bring yet further advantages because they are less vulnerable to flow-induced surface i.e. hydroelastic instabilities52.) But boundary-layer stability theory is based on the approximation of a spatially homogeneous flow (the so-called parallel-flow approximation) which perforce has to also assume an infinitely long compliant wall. An obvious question therefore arises. Do panels as short as a few, or even one, T/S wavelengths retain their capability to suppress T/S waves? This question has been investigated in detail by means of numerical simulation by Davies and Carpenter39,40. Figure 6 is based on one of these simulations. As discussed earlier, it can be seen that the compliant panel exhibits a complex response. It is typical of a case where the frequency of the T/S wave is above the cutoff frequency of the panel. The spectrum displayed in Figure 6 a shows that the response is a superposition of the incident T/S wave and two wall-based eigenmodes. Despite this complex response, the compliant panel in Figure 6 does suppress the growth of T/S waves because they emerge from the panels trailing edge with a muchreduced amplitude. In many respects Figure 6 depicts a rather extreme example. No wall damping was included and hinged end conditions were assumed at the leading and trailing edges. The use of more realistic clamped end conditions considerably reduces the extreme behaviour at the panels leading edge. Including light wall damping also reduces the complexity of the response. From our studies it appears that compliant panels as short as one T/S wavelength remain effective at suppressing T/S waves. This strongly suggests that multiplepanel compliant walls can maintain laminar flow indefinitely.
Figure 6. Numerical simulation of a Tollmien-Schlichting wave propagating in a laminar flat-plate boundary layer over a finite compliant panel of plate-spring type. a, Power spectrum obtained by analysing the instantaneous profile of wall displacement; the vertical broken lines correspond to eigenmodes of the coupled OrrSommerfeld/compliant-wall eigenproblem. b, Instantaneous spatial variation of the streamwise velocity perturbation plotted along and normal to the wall.
Although further experimental investigation remains desirable, the most important gap in knowledge concerns the effects of compliant walls on receptivity mechanisms, particularly those involving the generation of boundary-layer disturbances by freestream turbulence and suspensions of particles. This is the subject of our current research.
1. Gray, J., J. Exp. Biol., 1936, 13, 10. 2. Fish, F. E. and Hui, C. A., Mammal Rev., 1991, 21, 181. 3. Fein, J. A., in Proceedings of the International Symposium on Seawater Drag Reduction, Newport, RI, USA, 1998, pp. 429 432. 4. Schlichting, H., Boundary Layer Theory, 7th edn, McGraw Hill, New York, 1979. 5. Aleev, Y. G., Nekton, Junk, The Hague, 1977. 6. Kramer, M. O., J. Aero. Sci., 1957, 24, 459. 7. Kramer, M. O., J. Am. Soc. Nav. Engrs., 1960, 74, 341348. 8. Kramer, M. O., Adv. Hydrosci., 1965, 2, 111. 9. Babenko, V. V., Gintetskii, N. A. and Kozlov, L. F., Bionika, 1969, 3, 12. 10. Babenko, V. V. and Surkina, R. M., Bionika, 1969, 3, 19. CURRENT SCIENCE, VOL. 79, NO. 6, 25 SEPTEMBER 2000
ACKNOWLEDGEMENT. Our work on the hydrodynamics of compliant walls is supported by the UK Defence and Evaluation Research Agency and Engineering and Physical Sciences Research Council.
765